Background
The seroprevalence of human cytomegalovirus (HCMV) in the United States general population is approximately 60% and is even higher in certain socioeconomic groups [
1]. HCMV causes severe disease when immunity is suppressed such as in organ transplant recipients or during the later stages of HIV-1 infection [
2]. HCMV infection can be transmitted by bodily fluids of infected individuals, including saliva, blood, semen and cervical/vaginal secretions [
3]. Infection of infants can occur from exposure to genital fluids during birth and this route of infection can lead to mild to severe neurological sequelae.
Several studies suggest that alterations of genital tract flora in women can affect either initial infection by HCMV or virus replication/shedding. For example, HCMV DNA was detected more frequently in vaginal washings from women with bacterial vaginosis (BV) than in women with normal genital tract flora [
4]. BV is an alteration of the female genital tract flora consisting of an increase in both gram negative and gram positive bacteria [
5]. Increased HCMV shedding is also associated with concurrent
Chlamydia trachomatis or
Neisseria gonorrhoeae infection [
6]. Further, infection with
Trichomonas vaginalis,
N. gonorrhoeae, and BV are associated with increased intrauterine transmission of HCMV [
7]. The cause of the relationship between HCMV, BV and other sexually transmitted infections (STI) is not currently understood although inflammatory changes caused by STI could influence HCMV infection. Inflammation in genital tract infections is in many cases caused by the activation of genital tract cells through Toll-like receptor (TLR)-ligands derived from the pathogens;
N. gonorrhoeae,
T. vaginalis and BV flora all have been found to express products that activate TLR [
8‐
10].
In contrast to the studies that show enhancement of HCMV infection or shedding by genital tract infections, other studies show that stimulation through TLR can induce an antiviral state in cells or in animals [
11]. For example, replication of HSV-2 in vaginally-infected mice was prevented by intra-vaginal application of purified TLR ligands [
12,
13]. Similarly, intravenous injection of ligands for TLR3, -4, -5, -7, and -9 inhibit virus replication in Hepatitis B-transgenic mice [
14]. The anti-viral effect in these studies was mediated by induction of type I interferons via TLR stimulation [
14,
15].
In this study, we determined the effect of defined TLR ligands on HCMV replication as a model to better understand how changes in genital tract flora may enhance or inhibit HCMV replication in vivo. Since there are currently no animal models that are susceptible to HCMV, and only certain human cells are susceptible to HCMV infection, the effect of TLR ligands on replication of HCMV was assessed in foreskin fibroblasts (HFF), a previously described in vitro model of HCMV infection [
16,
17]. The TLR ligand effects were also studied in ectocervical tissue explants since HCMV was recently shown to replicate in this tissue and this may represent a model that more accurately represents in vivo infection by the virus [
18].
Materials and methods
Cells, tissues, & reagents
Human Foreskin Fibroblasts (HFF) were maintained in culture medium comprised of Minimum Essential Medium (Gibco, Carlsbad, CA) with 10 mM HEPES, 2 mM L-glutamine, 50 μg/ml gentamycin, 2.5 μg/ml amphotericin B, and 10% fetal bovine serum (FBS; BioWhittaker, Walkersville, MD). Cervical tissue was obtained at Northwestern University Medical Center from women undergoing planned hysterectomy for benign disease who had no history of cervical dysplasia. Patient consent was obtained by the treating physicians.
An HCMV clinical strain was engineered to express the Renilla green fluorescent protein under the control of the HCMV major immediate early promoter [
19]. The recombinant strain, HCMVPT30-gfp, produces extracellular virus and has similar growth kinetics as the parental strain [
18].
Purified lipoteichoic acid (LTA) from S. aureus and lipopolysaccharide (LPS) from E. coli O11:B4 were obtained from Sigma Aldrich (St. Louis, MO). Poly I:C (PIC) was obtained from Amersham (Piscataway, NJ). CpG 2395, a type C oligodeoxynucleotide, was generously contributed by Coley Pharmaceuticals (Wellesley, MA).
Treatment and infection of HFF
HFF were grown to 95% confluency in 24-well culture plates and treated with either medium alone or TLR ligands. After 24 h, medium was removed and assayed for cytokines. Cells were washed and CMVPT30-gfp was added (moi = 0.05). Cells were cultured for four hours, the virus inoculum was removed, and cells were cultured an additional 10 days. Monolayers were inspected by epifluorescent microscopy and the number of GFP-positive cells or clusters of cells (foci) was determined.
Cytokine ELISA
IL-8, IL-10, IL-12, and TNF-α were quantitated in cell culture fluids using CytoSet ELISA kits from Biosource (Carlsbad, California). IFN-α was tested using the IFN-α Module Set from Bender Medsystems (Burlingame, CA). IFN-β was assayed by coating 96-well flat bottom plates (NUNC, Rochester, NY) with 3 μg/ml monoclonal mouse anti-human IFN-β(Chemicon, Temecula, CA). Wells were blocked with 1% bovine serum albumin in phosphate buffered saline for 2 h at 25°C, washed three times and samples added and incubated for 1 h at 25°C. After washing, 3.5 μg/ml polyclonal rabbit anti-human IFNβ (Chemicon) was incubated in wells for one hour at 25°C followed by a 1/10,000 dilution of mouse anti-rabbit coupled to horseradish peroxidase (Chemicon) for 1 h at 25°C.
IFN-β neutralization
HFF were grown to 95% confluency in 24-well culture plates and treated with either medium alone, Poly I:C (10 μg/ml) or LPS (10 μg/ml) for 24 h. Cells were washed twice with medium and incubated at 37°C in 1 ml of fresh medium for 1 h to maximize removal of residual stimuli. Medium was replaced with 1 ml of complete medium and cells were cultured for an additional 24 hour period. Conditioned supernatants were collected and incubated with either complete medium, rabbit polyclonal anti-IFN-β neutralizing antiserum (Chemicon) (final concentration of 2 × 104 neutralization units/ml), or normal rabbit serum (NRS) (diluted 1:500 to give the same concentration of rabbit antibody) for 1 h at 37°C. The treated supernatants were then transferred to fresh 24-well culture plates containing confluent naïve HFF, and cultured for 24 hours. Conditioned medium was then removed and HFF challenged with CMVPT30-gfp. Fluorescent cells were counted by microscopy on day 10 after infection.
Treatment and infection of ectocervical tissue
Cervical tissues were washed extensively, cut into pieces of approximately 3 mm
3, and cultured in 48 well plates similar to a previously described method [
20] except that three ectocervical tissue pieces were cultured in each well of 48 well plates [
18]. Tissues were cultured in 0.5 ml medium containing Dulbeco's Modified Essential Medium, 24% Ham's nutrient mixture, 5 μg/ml insulin, 50 μg/ml gentamicin, 100 U penicillin/100 μg/ml streptomycin, 20 mM HEPES, 2 mM L-glutamine, 1 mm sodium pyruvate, and 10% FBS. TLR ligands were added to wells and cultured for 24 hours. Culture supernatants were removed and assayed for cytokines. Tissue pieces were washed and infected with HCMV (10
5 pfu per well) for four h at 37°C. Tissue pieces were washed again and then cultured for 10 days.
PCR quantitation of HCMV infection
Ectocervical explant tissue samples that were infected with HCMV were harvested and weighed. DNA was extracted using the Qiamp DNA Mini kit (Qiagen, Valencia, CA) and assayed by real-time PCR using primers for the DNA Polymerase gene of HCMV [
18]. The forward primer used was 5'-CTCGTGCGTGTGCTACGAGA-3' and the reverse primer used was 5'-GCCGATCGTRAAGAGATGAAGAC-3'. A FAM-AGTGCAGCCCCGRCCATCGTTC-TAMRA probe was used for detection of amplified product and a standard curve was generated using known copy numbers of genomic DNA from HCMV strain AD169 (Advanced Biotechnologies Inc., Columbia, MD). Results were expressed as HCMV copies/mg tissue.
Expression of TLR by HFF and ectocervical explant tissue
HFF, tissue, HEK293 were lysed and RNA extracted using the RNeasy Mini Kit (Qiagen, Stanford Valencia, CA). cDNA was made from 1 μg RNA from each cell type using the RT-PCR Kit from Clontech (Palo Alto, CA). The TLR primers were designed using Clone Manager Primer Software (4 Sci-ed, Durham, NC) based on gene sequences obtained from GeneBank (National Center of Biotechnology Information, NIH, Bethesda, MD). The primers were; TLR2 (F 5-CTCCAATCAGGCTTCTCT-3, R 5-TCAGTATCTCGCAGTTCC-3); TLR3 (F 5-GCATTCGGAATCTGTCTCTG-3, R 5-ATTCCTGGCCTGTGAGTTCT-3); TLR4 (F 5-GATGCCAGGATGATGTCT-3, R 5-CCGCAAGTCTGTGCAATA-3); TLR9 (F 5-TACCTTGCCTGCCTTCCTAC-3, R 5-CAACACCAGGCCTTCAAGAC-3); and GAPDH (F 5-GAAGGTGAAGGTCGGAGTC-3, R 5-GAAGATGGTGATGGGATTTC-3). Amplification was carried out using a GeneAMP Thermocycler (Perkin Elmer, Norwalk, CT) with a thermocycler profile as follows; Stage 1, 94°C (5 min); stage 2, 35 cycles of 94°C (45 sec), 62°C (45 sec) and 72°C (1 min) and Stage 3, 72°C (10 min).
Discussion
Sexually transmitted microbial diseases or bacterial vaginosis expose genital tract cells to TLR ligands. In this study we performed experiments to determine if exposure to defined TLR ligands affects HCMV infection and found that TLR ligands inhibit HCMV infection of both HFF and ectocervical explant tissue through induction of IFNβ. While no previous studies directly investigated the effect of TLR ligand stimulation of cells in vitro on HCMV infection, Sainz et al [
25] showed that the pretreatment of HFF with either IFN-α, IFNβ, or IFN-γ inhibited HCMV infection. Several previous studies showed induction of IFNβ in HFF, HEK fibroblasts, and human lung fibroblasts in response to stimulation with Poly I:C [
26‐
28].
While the effect of genital microbial infections on initial HCMV infection of women has not been reported, Ross et al. [
4] recently reported that HCMV shedding was found at a higher rate in women with BV than in women with normal flora. Infection with
T. vaginalis, gonorrhea, and BV were independently associated with intrauterine transmission of HCMV [
7]. Thus, these clinical studies show that under some in vivo conditions, HCMV infection can be enhanced by infections with other infectious agents. This suggests that TLR ligands may enhance HCMV infection in vivo since GC,
T. vaginalis and BV all have TLR ligands (TLR2, TLR4 and TLR2 respectively) associated with their infections [
8‐
10]. The clinical studies contrast with the findings of our in vitro and ex vivo studies where inhibition by defined TLR ligands was observed. A possible explanation for the differences could be that many of the clinical infections are chronic infections that in vitro 24 and 48 hour treatments with TLR ligands fail to accurately model. Also, in vivo adaptive immune responses or other stimuli may be present that affect HCMV that are lacking in vitro. Further studies are needed to understand these apparent differences.
A recent study showed that during infection with murine CMV, virus replicates to higher levels in mice lacking TLR2 [
29]. Depletion of Natural Killer (NK) cells eliminated the difference between TLR2-positive and TLR2-negative mice suggesting NK cells were involved in virus suppression in TLR2-positive mice. Also, type 1 interferon was lower in the TLR2 negative mice suggesting a role in virus suppression. The CMV inhibition in mice is different than the in vitro HCMV inhibition described in our study since in the mice no exogenous TLR ligands were given before infection. Intact HCMV virions have been reported to activate TLR2, possibly via glycoproteins B and H [
30,
31], although murine CMV is not known to have this activity. Iverson et al. [
32] showed that human NK cells can suppress HCMV through secretion of IFNβ, and NK cells can be stimulated through certain TLR including TLR2 [
33]. In our in vitro studies, no NK cells were present in HFF cultures showing that TLR3- and TLR4-ligands had a direct effect on the HCMV infection targets. However, in ectocervical tissue, it is possible that targets of HCMV infection as well as non-targets, such as immune cells, could have produced interferons. In mice, murine HCMV replicates to higher levels in mice deficient in TLR9 or MyD88 [
34,
35]. This higher replication is again associated with lower levels of type 1 IFN and decreased NK cell activity. However, mouse embryonic fibroblasts, dendritic cells and macrophges, and human fibroblasts have all been shown to secrete IFNβ in response to stimulation with LPS [
22,
36‐
38]. Thus, it is likely that multiple cell types in ectocervical tissues secrete IFNβ and contribute to the anti-HCMV effect.
Another interesting observation made in the current study was that the response pattern to the TLR ligands was different between HFF and ectocervical explants. Anti-HCMV responses in HFF were only found with TLR4 and TLR3 ligands while significant HCMV inhibition was induced by ligands to TLR2, TLR3, TLR4 and TLR9 in ectocervical explants. In our studies, IL-8 was measured to determine the responsiveness of HFF and explants to the TLR ligands. The IL-8 response pattern to the TLR ligands was also different between HFF and ectocervical explants with only TLR3 and TLR4 ligands inducing IL-8 in HFF but TLR3, TLR4 and TLR9 ligands inducing IL-8 in the tissue. Analysis of mRNA indicated that the ectocervical tissue expressed all four of the TLR while HFF only expressed TLR3 and TLR4. Many cell types express restricted repertoires of TLR receptors. For example, many epithelial cells have been observed to lack expression of TLR4 but to respond to TLR2 ligands [
39]. This highlights the importance of using models to study HCMV infection that most closely mirror the types of cells that are present in vivo. Cultures of ectocervical tissue have been used to study factors that affect HIV-infection [
20] and to assess the interactions of HIV with HCMV [
18], but this is the first study to investigate how TLR ligands affect HCMV infection in this tissue.
The inability of a TLR9 ligand to inhibit HCMV in HFF may be due to a lack of expression of TLR9 in these cells. TLR2 is not generally recognized to activate signaling pathways that lead to IFN production and may explain the lack of anti-HCMV effect in HFF due to this TLR ligand [
40]. However, TLR2 induced an anti-HCMV effect in ectocervical tissue and this appeared to be dependent on IFNβ. The mechanism for induction of IFNβ by TLR2 in tissues is not known although as mentioned above, some cells may produce IFN in response to TLR2 ligands. Also, stimulation through TLR2 can upregulate a number of molecules involved in anti-viral responses such as TRIF [
41] possibly leading to enhanced IFN production by cells due to other stimuli.
In conclusion this study shows that defined TLR ligands inhibit HCMV replication via IFNβ which suggests that different types of flora in the female genital tract can influence HCMV infection. This further suggests that reactivation and shedding of HCMV in the genital tract may be determined by alterations in the normal flora, which results from underlying conditions such as bacterial vaginosis or sexually transmitted diseases.
Authors' contributions
SH performed all of the cultures experiments and participated in writing of the manuscript. NL obtained and processed cervical tissue and provided direction to the studies. MRZ performed the TLR expression studies. GTS provided overall direction and co-wrote the manuscript. All authors read and approved the final manuscript.